Add functionality of MPO with block diagonal form (#154)

* stack mpo

* add test

* add comments

* update

---------

Co-authored-by: JiaceSun <[email protected]>

Author: SUSYUSTC

renormalizer/mps/__init__.py

@@ -1,5 +1,5 @@
 from renormalizer.mps.backend import backend
-from renormalizer.mps.mpo import Mpo
+from renormalizer.mps.mpo import Mpo, StackedMpo
 from renormalizer.mps.mps import Mps, BraKetPair
 from renormalizer.mps.mpdm import MpDm
 from renormalizer.mps.thermalprop import ThermalProp, load_thermal_state

renormalizer/mps/gs.py

@@ -18,7 +18,7 @@
 from renormalizer.mps.matrix import multi_tensor_contract, tensordot, asnumpy, asxp
 from renormalizer.mps.hop_expr import  hop_expr
 from renormalizer.mps.svd_qn import get_qn_mask
-from renormalizer.mps import Mpo, Mps
+from renormalizer.mps import Mpo, Mps, StackedMpo
 from renormalizer.mps.lib import Environ, cvec2cmat
 from renormalizer.utils import Quantity, CompressConfig, CompressCriteria
 
@@ -45,14 +45,14 @@ def construct_mps_mpo(model, mmax, nexciton, offset=Quantity(0)):
     return mps, mpo
 
 
-def optimize_mps(mps: Mps, mpo: Mpo, omega: float = None) -> Tuple[List, Mps]:
+def optimize_mps(mps: Mps, mpo: Union[Mpo, StackedMpo], omega: float = None) -> Tuple[List, Mps]:
     r"""DMRG ground state algorithm and state-averaged excited states algorithm
 
     Parameters
     ----------
     mps : renormalizer.mps.Mps
         initial guess of mps. The MPS is overwritten during the optimization.
-    mpo : renormalizer.mps.Mpo
+    mpo : Union[renormalizer.mps.Mpo, renormalizer.mps.StackedMpo]
         mpo of Hamiltonian
     omega: float, optional
         target the eigenpair near omega with special variational function
@@ -67,7 +67,7 @@ def optimize_mps(mps: Mps, mpo: Mpo, omega: float = None) -> Tuple[List, Mps]:
     mps : renormalizer.mps.Mps
         optimized ground state MPS.
             Note it's not the same with the overwritten input MPS.
-    
+
     See Also
     --------
     renormalizer.utils.configs.OptimizeConfig : The optimization configuration.
@@ -95,14 +95,19 @@ def optimize_mps(mps: Mps, mpo: Mpo, omega: float = None) -> Tuple[List, Mps]:
         env = "L"
 
     compress_config_bk = mps.compress_config
-    
+
     # construct the environment matrix
     if omega is not None:
+        if isinstance(mpo, StackedMpo):
+            raise NotImplementedError("StackedMPO + omega is not implemented yet")
         identity = Mpo.identity(mpo.model)
         mpo = mpo.add(identity.scale(-omega))
         environ = Environ(mps, [mpo, mpo], env)
     else:
-        environ = Environ(mps, mpo, env)
+        if isinstance(mpo, StackedMpo):
+            environ = [Environ(mps, item, env) for item in mpo.mpos]
+        else:
+            environ = Environ(mps, mpo, env)
 
     macro_iteration_result = []
     # Idx of the active site with lowest energy for each sweep
@@ -111,7 +116,7 @@ def optimize_mps(mps: Mps, mpo: Mpo, omega: float = None) -> Tuple[List, Mps]:
     res_mps: Union[Mps, List[Mps]] = None
     for isweep, (compress_config, percent) in enumerate(mps.optimize_config.procedure):
         logger.debug(f"isweep: {isweep}")
-        
+
         if isinstance(compress_config, CompressConfig):
             mps.compress_config = compress_config
         elif isinstance(compress_config, int):
@@ -156,19 +161,19 @@ def optimize_mps(mps: Mps, mpo: Mpo, omega: float = None) -> Tuple[List, Mps]:
         for res in res_mps:
             res.compress_config = compress_config_bk
         logger.info(f"{res_mps[0]}")
-    
+
     return macro_iteration_result, res_mps
 
 
 def single_sweep(
     mps: Mps,
-    mpo: Mpo,
+    mpo: Union[Mpo, StackedMpo],
     environ: Environ,
     omega: float,
     percent: float,
     last_opt_e_idx: int
 ):
-    
+
     method = mps.optimize_config.method
     nroots = mps.optimize_config.nroots
 
@@ -210,18 +215,26 @@ def single_sweep(
         if omega is None:
             operator = mpo
         else:
+            assert isinstance(mpo, Mpo)
             operator = [mpo, mpo]
 
-        ltensor = environ.GetLR("L", lidx, mps, operator, itensor=None, method=lmethod)
-        rtensor = environ.GetLR("R", ridx, mps, operator, itensor=None, method=rmethod)
+        if isinstance(mpo, StackedMpo):
+            ltensor = [environ_item.GetLR("L", lidx, mps, operator_item, itensor=None, method=lmethod) for environ_item, operator_item in zip(environ, operator.mpos)]
+            rtensor = [environ_item.GetLR("R", ridx, mps, operator_item, itensor=None, method=rmethod) for environ_item, operator_item in zip(environ, operator.mpos)]
+        else:
+            ltensor = environ.GetLR("L", lidx, mps, operator, itensor=None, method=lmethod)
+            rtensor = environ.GetLR("R", ridx, mps, operator, itensor=None, method=rmethod)
 
         # get the quantum number pattern
         qnbigl, qnbigr, qnmat = mps._get_big_qn(cidx)
         qn_mask = get_qn_mask(qnmat, mps.qntot)
         cshape = qn_mask.shape
 
         # center mo
-        cmo = [asxp(mpo[idx]) for idx in cidx]
+        if isinstance(mpo, StackedMpo):
+            cmo = [[asxp(mpo_item[idx]) for idx in cidx] for mpo_item in mpo.mpos]
+        else:
+            cmo = [asxp(mpo[idx]) for idx in cidx]
 
         use_direct_eigh = np.prod(cshape) < 1000 or mps.optimize_config.algo == "direct"
         if use_direct_eigh:
@@ -285,15 +298,15 @@ def single_sweep(
     return micro_iteration_result, res_mps, mpo
 
 
-def eigh_direct(
+def get_ham_direct(
     mps: Mps,
     qn_mask: np.ndarray,
-    ltensor: xp.ndarray,
-    rtensor: xp.ndarray,
+    ltensor: Union[xp.ndarray, List[xp.ndarray]],
+    rtensor: Union[xp.ndarray, List[xp.ndarray]],
     cmo: List[xp.ndarray],
     omega: float,
 ):
-    logger.debug(f"use direct eigensolver")
+    logger.debug("use direct eigensolver")
 
     # direct algorithm
     if omega is None:
@@ -347,6 +360,23 @@ def eigh_direct(
             )
             ham = ham[:, :, :, :, qn_mask][qn_mask, :]
 
+    return ham
+
+
+def eigh_direct(
+    mps: Mps,
+    qn_mask: np.ndarray,
+    ltensor: Union[xp.ndarray, List[xp.ndarray]],
+    rtensor: Union[xp.ndarray, List[xp.ndarray]],
+    cmo: List[xp.ndarray],
+    omega: float,
+):
+    if isinstance(ltensor, list):
+        assert isinstance(rtensor, list)
+        assert len(ltensor) == len(rtensor)
+        ham = sum([get_ham_direct(mps, qn_mask, ltensor_item, rtensor_item, cmo_item, omega) for ltensor_item, rtensor_item, cmo_item in zip(ltensor, rtensor, cmo)])
+    else:
+        ham = get_ham_direct(mps, qn_mask, ltensor, rtensor, cmo, omega)
     inverse = mps.optimize_config.inverse
     w, v = scipy.linalg.eigh(asnumpy(ham) * inverse)
 
@@ -360,14 +390,13 @@ def eigh_direct(
     return e, c
 
 
-def eigh_iterative(
+def get_ham_iterative(
     mps: Mps,
     qn_mask: np.ndarray,
-    ltensor: xp.ndarray,
-    rtensor: xp.ndarray,
+    ltensor: Union[xp.ndarray, List[xp.ndarray]],
+    rtensor: Union[xp.ndarray, List[xp.ndarray]],
     cmo: List[xp.ndarray],
     omega: float,
-    cguess: List[np.ndarray],
 ):
     # iterative algorithm
     method = mps.optimize_config.method
@@ -428,6 +457,34 @@ def eigh_iterative(
     # contraction expression
     cshape = qn_mask.shape
     expr = hop_expr(ltensor, rtensor, cmo, cshape, omega is not None)
+    return hdiag, expr
+
+
+def func_sum(funcs):
+    def new_func(*args, **kwargs):
+        return sum([func(*args, **kwargs) for func in funcs])
+    return new_func
+
+
+def eigh_iterative(
+    mps: Mps,
+    qn_mask: np.ndarray,
+    ltensor: Union[xp.ndarray, List[xp.ndarray]],
+    rtensor: Union[xp.ndarray, List[xp.ndarray]],
+    cmo: List[xp.ndarray],
+    omega: float,
+    cguess: List[np.ndarray],
+):
+    # iterative algorithm
+    inverse = mps.optimize_config.inverse
+    if isinstance(ltensor, list):
+        assert isinstance(rtensor, list)
+        assert len(ltensor) == len(rtensor)
+        ham = [get_ham_iterative(mps, qn_mask, ltensor_item, rtensor_item, cmo_item, omega) for ltensor_item, rtensor_item, cmo_item in zip(ltensor, rtensor, cmo)]
+        hdiag = sum([hdiag_item for hdiag_item, expr_item in ham])
+        expr = func_sum([expr_item for hdiag_item, expr_item in ham])
+    else:
+        hdiag, expr = get_ham_iterative(mps, qn_mask, ltensor, rtensor, cmo, omega)
 
     count = 0
 

renormalizer/mps/mpo.py

@@ -127,7 +127,7 @@ def ph_onsite(cls, model: HolsteinModel, opera: str, mol_idx:int, ph_idx=0):
     def intersite(cls, model: HolsteinModel, e_opera: dict, ph_opera: dict, scale:
             Quantity=Quantity(1.)):
         r""" construct the inter site MPO
-        
+
         Parameters
         ----------
         model : HolsteinModel
@@ -142,7 +142,7 @@ def intersite(cls, model: HolsteinModel, e_opera: dict, ph_opera: dict, scale:
         Note
         -----
         the operator index starts from 0,1,2...
-        
+
         """
 
         ops = []
@@ -330,7 +330,7 @@ def apply(self, mp: MatrixProduct, canonicalise: bool=False) -> MatrixProduct:
         # todo: use meta copy to save time, could be subtle when complex type is involved
         # todo: inplace version (saved memory and can be used in `hybrid_exact_propagator`)
         # the model is the same as the mps.model
-        
+
         assert self.site_num == mp.site_num
         new_mps = self.promote_mt_type(mp.copy())
         if mp.is_mps:
@@ -388,14 +388,14 @@ def apply(self, mp: MatrixProduct, canonicalise: bool=False) -> MatrixProduct:
 
     def contract(self, mps, algo="svd"):
         r""" an approximation of mpo @ mps/mpdm/mpo
-        
+
         Parameters
         ----------
         mps : `Mps`, `Mpo`, `MpDm`
         algo: str, optional
             The algorithm to compress mpo @ mps/mpdm/mpo.  It could be ``svd``
-            (default) and ``variational``. 
-        
+            (default) and ``variational``.
+
         Returns
         -------
         new_mps : `Mps`
@@ -476,3 +476,16 @@ def is_hermitian(self):
     def __matmul__(self, other):
         return self.apply(other)
 
+
+class StackedMpo:
+    """
+    An effective sparse representation of MPO in the block diagonal form.
+    When it enters into the optimization, the Hamiltonian is calculated as
+    the sum of Hamiltonians generated by each MPO, then the Hamiltonian is
+    diagonalized and the MPS is updated.
+
+    Usage:
+    optimize_mps(mps, StackedMpo([mpo1, mpo2, ...]))
+    """
+    def __init__(self, mpos: List[Mpo]):
+        self.mpos = mpos

renormalizer/mps/tests/test_gs.py

@@ -9,7 +9,7 @@
 from renormalizer.model import Model, h_qc
 from renormalizer.mps.backend import primme
 from renormalizer.mps.gs import construct_mps_mpo, optimize_mps
-from renormalizer.mps import Mpo, Mps
+from renormalizer.mps import Mpo, Mps, StackedMpo
 from renormalizer.tests.parameter import holstein_model
 from renormalizer.utils.configs import OFS
 from renormalizer.mps.tests import cur_dir
@@ -136,3 +136,14 @@ def test_qc(with_ofs):
     print(mpo)
     gs_e = min(energies)
     assert np.allclose(gs_e, fci_e, atol=5e-3)
+
+
+def test_stackedmpo():
+    scheme = 1
+    method = '1site'
+    mps, mpo = construct_mps_mpo(holstein_model.switch_scheme(scheme), procedure[0][0], nexciton)
+    mps.optimize_config.procedure = procedure
+    mps.optimize_config.method = method
+    energies1, _ = optimize_mps(mps.copy(), mpo)
+    energies2, _ = optimize_mps(mps.copy(), StackedMpo([mpo, mpo]))
+    assert np.all(np.abs(np.array(energies2) - np.array(energies1) * 2) < 1e-8)